AI search changes the route to a recommendation, not the need for a useful website and a credible reputation. On Practical AI, Discovered Labs co-founders Liam Dunne and Ben Moore argue that businesses should distinguish what an answer system retrieves, what it visibly cites, and whether a buyer ultimately acts: those are different outcomes. Their Reddit example is the clearest illustration—material can help ground an answer while another source receives the displayed citation, so fewer citations need not mean less influence. Conversational buyers also supply detailed constraints, creating opportunities for specific information about pricing, integrations, and product fit rather than another generic keyword page. Dunne’s framework is relevancy, consensus, and consistency: explain the offer clearly, earn corroboration outside the company’s own site, and remain recognizable across channels. Moore’s strongest advice is to measure variation before claiming progress; one favorable prompt or a moving dashboard is not a controlled experiment. The caveat is that these are agency practitioners’ observations, not an independent audit of proprietary search systems: their claims about shared internal algorithms, training schedules, and text fingerprinting require separate evidence. The useful next step is less glamorous and more durable—publish accurate information, test realistic buyer questions repeatedly, and check whether agents can complete the next action, not merely find a paragraph.
“a lot of people waste a lot of marketing hours and a lot of engineering time around things that aren’t statistically significant.” — Ben Moore, 14:52
“So I think it’s relevancy, consensus, consistency.” — Liam Dunne, 38:50
Based on the full publisher transcript; quotations checked against its timestamped captions.